Subscribe
Share
Share
Embed
In the Original Thinking Podcast, experts and academic colleagues discuss their latest research and original thinking at Alliance MBS.
For a list of our latest webinars, news and useful business content please visit ambs.ac.uk
JIM PENDRILL:
Good afternoon, everybody. A very warm welcome to our latest Original Thinkers podcast here at Alliance Manchester Business School. I'm Jim Pendrill, I'm Research Communications lead here at the school. I'm delighted, for our latest podcast, to be joined by Erik Beulen, who is Professor of Information Management here at the school. Afternoon, Erik. Nice to see you.
ERIK BEULEN:
Thank you.
JIM PENDRILL:
Today's topic is AI. But Erik has been looking at AI in sort of perhaps many different ways over the last many years, really. And although we, every day, hear about AI and the amazing advances it's making, Erik has been writing for many years around the ethics of AI. And in fact, I'm going to start with a quote that you gave me about a year ago, actually, for an article that we put together and you said, that AI needs to be lawful, explainable, transparent and responsible. And I'm going to start with just asking you, do you think that's achievable?
ERIK BEULEN:
It must be. So, it's not a matter of could it or could it not? Organisations need to be very diligent on that. And why? Because there's also, if you don't do that, a potential risk of doing damage, negative things to your organisation. So you need to be very diligent. Put that effort in it. So if you can't comply with these requirements then don't use AI or use AI in a different way. Less complex. Just break it up in different types of algorithms and then build that up to the complexity that you think you need to improve your service, to improve your product. And then you can use this type of AI. So simplicity is I think, helpful in meeting these four criteria.
JIM PENDRILL:
But do you think enough leaders of businesses and organisations have the skills required to ensure that AI remains, you know, lawful, explainable, transparent and responsible? And to what extent, as well, do governments and regulators have a role to play here?
ERIK BEULEN:
Well, that's lots of questions. Let me try to...
JIM PENDRILL:
Think it out. Yeah. Yeah.
ERIK BEULEN:
Let me try to address a couple of things first. So I think we need to distinguish between the teams that are building the algorithms. So these are the data scientists together with the data engineers. They are more let's say on, OK, well, how can we use the data that we have? And they see lots of opportunities and they see ways that they can use the data, which is meaningful for an organisation. But then it comes down to leadership.
Then it comes down to governance to really make sure that what is being implemented in an organisation, that is meeting these four criteria. And that's in terms of where you need to have lots of checks and balances. What are we doing? And let me give an example of what we came across in our research, is that you have a team, and a team has a set of guidelines, rules where they need to adhere to. So they need to score themselves.
If there is a concern then that's been reviewed one level up. If there is a concern then that's reviewed one level up and that's the way to do it. Because let's be honest, the complexity, the sets of data that organisations are dealing with are growing by the day. So if you want to be in control, if you really want to make sure that you're doing what is right for the organisation, then you need to have that type of monitoring all along the way. And is that simple? Is that straightforward?
I would say no, but is that necessary? I would say yes. And that's I think, where organisations that are putting a lot of effort in that will, in the long run, make the difference compared to companies which are less diligent on that one.
JIM PENDRILL:
From your sort of experience, do you think most organisations now, or leaders of organisations understand that and are implementing those kind of structures? Or are we still at early days here? And a lot of people just don't quite know what to do in terms of managing these data and these algorithms?
ERIK BEULEN:
I think it's getting better. And I think one of the reasons it's getting better is that there is a lot of legislation currently implemented here in Europe, but also in other parts of the world. We got the AI Act here in the EU. Here. We got also in the US legislation that is specifically focusing on this and helping organisations to be mindful of that and to give structure. And I think that is what it is. It's a combination of understanding, being part of the DNA of an organisation. And on the other hand, legislation that is supporting to achieve meeting these criteria.
JIM PENDRILL:
OK. Well, I'm sure these are these are issues which you cover in your new book, which you're also going to talk about, called 'Data Analytics and Digital Transformation', which you co-wrote with Marla Dans, who is the head of Data Management and Governance at the Chicago Trading Company. Just tell us a little bit about the background to the book. What made you both write it and what are the key messages within the book?
ERIK BEULEN:
Yeah, sure, sure. Yeah. So, well, first of all, I would like to explain that when I'm writing a book, I'm always looking for a co-author, which is very complementary to my skills. And that is what I found in Marla, in terms of her wealth of industry experience and bringing that into the book, together with also her work that we did together with TCS, which is a large service provider, which brought in subject matter experts, use cases and that type of thing.
So that is in terms of what we would like to achieve also with the book, make it into something which is useful for organisations to think, OK, well, this is what we're going to do in terms of digital transformation combined with data," because what we basically, Marla and I both see as what is important, that if you want to be successful in digital transformations, then you need to leverage your data.
You need to leverage data analytics because that is really driving data-driven leadership, that is driving companies forward, combining with what you have in terms of products or services with data, and then just make it better. And that is why, yeah, we were highly motivated to, um, yeah, to write this book.
JIM PENDRILL:
And in terms of the key messages, Erik, I mean, as well as leveraging your data, what are the other key points you make, you make in the book?
ERIK BEULEN:
Yeah. So what is important is, is that you fully embrace that. So when we're talking about digital transformations then we talk about the role of the Chief Digital officer. And our point of view is it's important to have a person that is responsible for that. But basically, it would be even better if that is in the DNA of your organisation. So if you then think about capabilities, if you then think about corporate development, then that is something which is important, to have that.
A lot of organisations are on that journey. I would say digital transformations are there, let's say 2012, 2013. So we're more than ten years on its way. So really leading organisations, they have that. If you look to the data part then we see there's something similar. So we see organisations appointing chief data officers, both CDOs but different in terms of nature, digital versus data. And what we see there is something similar. Also there, it's important that there is a person who is basically setting this up, implementing this and getting it off the ground.
And then where a lot of organisations are right now is that they're building data ownership, and data ownership can only be in the business. So if you are responsible for a department, for a division, then you need to own the data. And owning the data means also taking care and responsibility for data quality and taking care and responsibility for who can access the data and that type of things. And that is where also another key point of our book is. If we're talking about access, we can talk about, OK, well, this role, this person, this department, they should have access to the data.
But if we look to where digital transformations are really successful, then it's not limited to the organisation, whether that's a private organisation or a public organisation, but it's in the value chain. And if you then want to share data, if you then want to collaborate, if you then want to co-create, then you need to be very, very strong in terms of how we're going to do that. So that's processes, appointments, that type of thing. So you really need to be strong there.
JIM PENDRILL:
I mean, you've written a lot about the link between companies and organisations and service providers.
ERIK BEULEN:
Yeah.
JIM PENDRILL:
Just for the purposes of the audience, just explain a little bit about what you mean by that and how important that is, actually.
ERIK BEULEN:
Um...
JIM PENDRILL:
If it's possible to explain that in just a few minutes.
ERIK BEULEN:
If you look to what is currently going on in terms of collaborating organisations, then having support of service providers is essential. And why? Because all the things that are non-core, that is what you would like to bring in. And that is helpful just to have a standardisation, to have easy interfacing across the value chain, but also the parts where you want to be innovative, OK, good. How are we going to do that?
If you're thinking about OK, well, how are we going to do a lot in terms of AI, can you have the resources in your own organisation to set that up? Is that something which is working, or is it better just to partner with a service provider, and that partnering is then different from the commodity part? Because then you just want price if you want to co-create. And then there is also exclusivity and you're having different KPIs on how to collaborate.
JIM PENDRILL:
Yeah. And this is really important because this is where the ethics comes in and making sure that you're, you know, you're abiding by all the regulations or, you know, governance etc.. I mean, this is, and I guess the danger can be that there can be gaps appear.
ERIK BEULEN:
Yeah. But that is in terms of really thinking through where you want to be in terms of strategy and how you want to deliver that in terms of ecosystems. And if I look to organisations, the organisations that have a very clear view on what they want to do and have a very clear view on how to achieve that, and that's why there's a connecting the dots. And for me, in that whole ecosystem, there is a definite place for service providers because they can bring in that expertise. They can far more easily scale up and help organisations to the next level.
JIM PENDRILL:
Mhm. OK. Let's talk about a few other things which you've been, I know you've been researching a lot and one is around the specific measures that organisations should implement in order to ensure the ethical use of data. Again, just briefly, what sort of things are we talking about? What sort of measures are we talking about there? And because we hear a lot about ethics and AI at the moment, don't we? So this is extremely topical. I mean, what sort of things have you been looking at?
ERIK BEULEN:
Yeah, I think a couple of things here. So there is, especially from governments, a tendency to say, "Well, we need to register that, it needs to be open and everybody needs to look at that." I'm not so keen on that. And the reason is that if you're thinking about competitive advantage, if you're thinking about, OK, well, how will things work, then some of the things you would like to keep close to your chest, and that is where you need to make sure that is not happening.
And that's where in terms of if you are not putting things in a register where you say, “OK, well, that's the general public, all of us see what we're doing is all right, that you have the right measures.” What I see organisations are building is basically a type of governance structure where there is really taking a look on the algorithms. Can we explain that? What is the type of data sets that we're using? Also there we see that a technology tooling is helping to identify biases. And that is in terms of not using AI only for the algorithm, but also using AI for control. And that is, I think, a good combination of an open register where everybody can look at, good governance and good tooling systems just to do that.
JIM PENDRILL:
Yeah. I'm glad you mentioned bias because I was going to ask you about that later, but let's talk about it now. And again, something that you've talked about before, we've talked about is, you know, is how you ensure employees and service providers remain neutral and unbiased. You know, how can you ensure that opinions don't interfere with data collection processes? And this is a huge subject. But again, are there simple things you can do to ensure that you try and reduce the likelihood of biases? I mean, someone said to me recently that you can't eradicate bias in AI. It's almost - but you're probably, you might disagree with that statement. But, yeah, what is your view?
ERIK BEULEN:
I'm definitely not disagreeing with that. I think it's about consciousness and being very conscious on where you run risk and then just dive into that part deeper. So if you're dealing, for example, with sensitive matters related to privacy, then that requires far more diligence than when you're dealing with, let's say, transactional data of a certain process. So that in terms of you have limited capacity to address bias, make sure that that limited capacity is in the areas which you identified, Well, that might be risky. And that's a good way of doing that.
And it's also in the culture. I think if you're having a culture where, hey, you're open and you're transparent, then that's more easy to do. You also were referring to what will that be if there will be employees from a supplier and employees from the own organisation? I think in that respect, it's more easy because the interest of an external service provider had to build something which is not biased is super high. Because imagine what would happen if a supplier builds a biased system. Look at the headlines, what were then going to happen. You don't want that. So that scrutiny that a service provider puts on top of, 'is this bias or not?' is, in my experience, much more rigour than internal organisations.
JIM PENDRILL:
Hmm. Interesting. So yeah. Another area again here, which we wanted to talk about was cybersecurity, which often is lumped in with AI, trust and society and ethics. But obviously extremely important. I mean, where does cybersecurity fit into this ecosystem you're describing?
ERIK BEULEN:
I would say, it's also at the heart and let me also explain why that's my view. If we look to what organisations are doing, then we see two things. Far more collaboration and far more exchange of information. And exchange of information means linked systems. And that's where you need to be very mindful, because if one of your partners in your ecosystem is the victim of a cyber-attack, how then can you be sure that it's not also in your systems? And so you need to put a lot of protection in that. And protection is monitoring, 24/7.
It's in terms of what you want to do in authentication. I don't want to go into the technical details, but the IT team, the tech team needs to put a lot in place. Also here we can see that organisations, service providers are leveraging artificial intelligence to do threat analysis. And that is in terms of what is going on in the monitoring important. But it's also a matter of looking ahead. And that is where I'm really looking. OK, well, hey, what is going on in this area? Identifying, take preventive measures. If you see things happening in other parts of the world, what are we going to do? Anticipating is the best way, way forward.
But it's a super dangerous threat. And potentially it can bring organisations down or at a minimum, have a big impact and we've seen that a lot.
JIM PENDRILL:
Oh yeah. Every day. We almost every week probably hear something, don't we?
ERIK BEULEN:
Unfortunately, yes.
JIM PENDRILL:
Yeah. But as you say, yeah, organisations, there are things, a lot of things they can do and they've got to place it at the heart of these systems.
ERIK BEULEN:
Yeah.
JIM PENDRILL:
Another thing I wanted to talk about was leadership. We've touched on it really actually a little bit. But you know, what leadership is needed to oversee a successful digital transformation and oversee the use of AI, all these debates around ethics and, you know, explainability of AI. What sort of characteristics leaders need? And I guess they're changing quite quickly, what those sort of characteristics are. I mean, you mentioned the chief data officers, but maybe what does the CEO need above that? What do they need? Or do they rely on the CD CDO to...
ERIK BEULEN:
They should not. They should not. Oh no, no, no, no, no. And I think that's what we're trying also to describe in the book, how we're speaking about digital transformation 1.0, 2.0 and 3.0. And I think if we're looking ahead in terms of what is important to capture, then that's basically that all of this innovation that that is part of, and thinking in terms of how we can do that, that's part of the DNA across leadership. I referred earlier to corporate development.
I think that is super important that you think in terms of what type of capabilities do we need, and how does that reflect on our culture and how are we going to deal with that? And that's why in terms of no longer needing a chief data, a chief digital officer, no longer needing a chief data officer because it's in the DNA. And so if you think about how can an organisation transform and have to transform, that's in that type of area.
And that brings organisations forward and the organisations who are not able to be on that path, they will be just overtaken by companies who are organisations who are.
JIM PENDRILL:
I mean, we're sat here today in a business school. I mean, you know, I guess business schools across the world are sort of thinking, "OK, what do we need to be teaching the leaders of tomorrow?" Well, what do these schools, per se, need to be teaching?
ERIK BEULEN:
Yeah, I think a couple of things. So you need leaders who have really subject matter expertise. But that, if you're thinking about true leadership, thinking about CEOs as a type of roles, that's more general leadership. And that's where I would like to come back to the DNA, because they need to understand, they need to be tech savvy. They need to understand the implications of legislation more than ever before. Why? Because they are personally on the hook for things like cybersecurity, for things like AI. And that's where regulations are creeping up towards them. And that is in terms of, yeah, understanding the real, yeah, impact of all of this and being able to oversee that and to drive an organisation into, yeah, a certain area. I think that's what is needed.
JIM PENDRILL:
You mentioned legislation. You mentioned it both in the EU and the US. I mean, the narrative you hear is one that it's struggling to catch up, but it sounds like they are catching... It is catching up, or is it not catching up quick enough right now? Or will it ever catch up, given the proliferation of AI algorithms?
ERIK BEULEN:
Yeah. And I think that's where I would like to distinguish between, let's say, EU regulation, who is trying to be very prescriptive. This is allowed. This is not allowed. Versus a more, let's say UK perspective on that in terms of OK, well this is what we want to achieve. So objective setting and then have a legal check or where is this breaching that objective or not? And so then that's more agnostic of what type of technology development there is. And I think that is more, far more let's say effective in terms of achieving what you want to achieve.
So and I think then that is where, how we need to find a little bit of middle ground. It needs to be specific enough in terms of this is what we mean, but it needs to also be open enough to say, OK, well, we hadn't seen this coming, but OK, well this is how we are dealing with that.
JIM PENDRILL:
Yeah. Yeah. So, broadly, are you quite positive that regulation is going in a good direction or are you... Do you think that, you know, battles are going to lie ahead to try and sort of... because there have been sort of countries coming together and talking about it, haven't they? The UK hosted that summit back, end of last year.
ERIK BEULEN:
Yeah.
JIM PENDRILL:
A lot of activity around this, you know.
ERIK BEULEN:
Yeah. So I think a couple of things, I think, that the international collaboration between governments and governmental organisations, so also Helsinki, United Nations and that type of organisations is very good. And that is helpful because there is an understanding that we got a lot of things that we need to take care for, and that's helpful in that collaboration. And it was good to see that also, a lot of EU countries were present in November in your conference here in your country to talk about how can we do cyber and cyber security better?
So that is, I think, very, very promising. And I think that type of alignment, that is a good, a good thing because if you look to important countries that were present there, that is a that is very, very good.
JIM PENDRILL:
Yeah. OK. So, perhaps just wrapping up, I'd like to go back to that first question that I asked you, which was, you know, that quote you gave that AI needs to be held lawful, explainable, transparent and responsible.
ERIK BEULEN:
Yeah.
JIM PENDRILL:
Do you think we're getting there? But we've got a lot, obviously got quite a way to go by sounds of what you're saying.
ERIK BEULEN:
Yeah. But by nature, I'm always optimistic and a positive person, but I think if we look to the facts, that's also substance by facts. If we see the much debate that there is on that, what companies are doing, what they're investing in that, and don't underestimate that, that is for me, all proof points that we're doing the right things. Hey, could we do better? Yeah, always. But that's if we're moving into the right direction, then that is something which I, yeah, I really appreciate and also think it's necessary.
JIM PENDRILL:
Yeah, yeah. And just finally, obviously, you just produced your latest book. What's your next book going to be about or have you not...
ERIK BEULEN:
Yeah. Sure. Sure. So, I think we talked about it already also in this podcast. And what we're looking for is just how can we give guidance to AI governance? Because that is, I think, a very emerging topic and also a very complex topic. So we're working again with industry partners just to see what we can bring forward on that. And I explained already a couple of ideas that we have on that. So, I'm looking forward to it.
JIM PENDRILL:
Yeah. Look forward to hearing about that in due course. Well, Erik, it's been great talking to you. It's been a fascinating discussion. A lot... We've covered a lot of ground in quite a short space of time. But I think we did deliberately, wanted to keep it not too technical today because it is complex, but yeah. And, look forward to hearing more in the future about your work. Thank you.
ERIK BEULEN:
Thank you. Thank you for having me, Jim. Thanks.
JIM PENDRILL:
So thanks, Erik, for those fascinating insights into the book. In the next section now, we're going to hear from Marla Dans from the Chicago Trading Company, who was a co-author on the book with Erik, who's going to share her own thoughts on why they put the book together and what its outcomes were.
MARLA DANS:
Hello, my name is Marla Dans. I'm a long-time veteran in chief data office, data management and data governance. I started in JP Morgan's chief data office ten years ago, and from there I went to a fintech firm called Tradeweb, which was fixed income trading platform where I built the programme from scratch. And then from Tradeweb I went to Chicago Trading, which is an options market maker, and I'm in Chicago today talking to you from this podcast.
So, why I wrote this book. So, a long time back, I had these, I had done so much work, and I had so much information that I wanted to put down on paper. And it's always been a long-time goal and a bucket list item for me to write a book. And when I was started in, in data analytics and data governance, we were in organisations that I noticed had full, you know, divisions that were dedicated to digital transformation. And then they were forming these full divisions that were dedicated to data management and governance, and they were calling them Chief Data Office.
So my thoughts were, how do these two new, brand new divisions in corporations work with each other? How does data analytics fuel digital transformation, and how does the data coming out of digital transformation, fuel data analytics? And that was the beginning and the beginning thoughts for this book. Then I met Professor Erik and we worked on some projects together. He had this whole experience and has written many books and has this whole academic perspective on data. So bringing the two worlds together really made for a very powerful book and a very interesting perspective on data analytics and digital transformation.
So it really, as I said, brings together a blend of that practical experience and academic rigour. The book offers original insights and perspectives. There's lots of survey results from industry experts, subject matter expert interviews on topics like ethics, real-life use cases where you can see the key, some key points in action. But the book really I would say is foundational, right? It has everything you need to get your data AI-ready. It includes data analytics and governance, tooling and processes, accountability with data stewards and data owners.
We talk about data quality, the process for identifying data quality issues as well as remediating them, which are two different things. And then going forward, we look to the increasing rigour around algorithmic assessments and the need for AI governance. So we hope you enjoy reading this book and stay tuned for more.